Sentinel Lymph Node Biopsy: Strategies forPathologic Examination of the Specimen

JOHN S. MEYER, MD1,2*1Department of Pathology, St. Luke’s Hospital, Chesterfield, Missouri

2Department of Pathology, Washington University School of Medicine, St. Louis, Missouri

Background and Objectives:Sentinel lymph node biopsy is a practicalsurgical procedure with high sensitivity for detection of metastases frombreast carcinoma in the axilla. It offers economy by avoidance of axillarydissection in the majority of breast carcinoma patients who have negativesentinel node biopsies, and provides an opportunity to study axillary mi-crometastases with high efficiency in a small volume of tissue.Methods: Sentinel node tissue is sliced at 2-mm intervals for fixation andparaffin embedding. Probabilities of finding spherical micrometastases ofspecific sizes randomly distributed in lymph nodes were calculated geo-metrically for several microsectioning plans.Results and Conclusions:Sentinel node tissue can studied by systematicserial sectioning technique designed to find metastases of given diameterswith specific probabilities. A procedure whereby three microsections areprepared repeatedly at intervals of 250mm appears to be practical. Twosections from each level can be examined by routine staining and the thirdby immunohistochemical stain; the latter is recommended particularly forinfiltrating lobular carcinoma. This method will find metastases of 0.25-mm diameter with theoretical probability of 1, and metastases of 0.10-mmdiameter with probability of 0.46, with reasonable costs. Metastases ofthese sizes are consequential and worth finding on biological and clinicalgrounds.J. Surg. Oncol. 1998;69:212–218. © 1998 Wiley-Liss, Inc.

KEY WORDS: metastases; sentinel lymph node biopsy; microsectioning

INTRODUCTION

Sentinel lymph node biopsy is a procedure based onthe concept that lymphatics from a given site make directconnection with only one or a few lymph nodes, and thatmetastasis from a tumor will appear first in these nodes.Injection of soluble dye or particulate matter into thetumor will demonstrate the lymph node or nodes thatprimarily drain the tumor. Among the nontoxic dyesavailable for this purpose are lymphazurin blue (isosul-fan blue) and fluorescein. Lymph nodes remain stainedby these two dyes for 30–60 min. Methylene blue stain-ing persists for up to 6 h, but methylene blue is contra-indicated in patients with glucose 6-phosphatase defi-ciency or unstable hemoglobins, and it may be toxic toepithelial cells [1]. Technetium-99m-labeled sulfur col-loid or colloidal albumin is useful for localization ofsentinel nodes by scintigraphic imaging or use of a hand-heldg-counter [2]. The half-life of the metastable Tc-99

isotope is 6.01 h. It decays by isomeric transition withg-ray emission into technetium 99, ab-emitter with ahalf-life of 213,000 years. Technetium can be injected1⁄2–3 h before scintiscanning of lymph node sites [3,4].The technetium method is more sensitive than dye, whichmay detect only two-thirds of nodes detected byg emis-sion after technetium uptake [5]. Gulec and associates [6]have reviewed the anatomic and physiologic rationale ofsentinel lymph node biopsy and note that a multicenterclinical trial is under way.

Micrometastases may be more readily found in senti-nel nodes than in standard axillary dissections, but thisimpression could be a result of more extensive search of

*Correspondence to: John S. Meyer, MD, St. Luke’s Hospital, 232 S.Woods Mill Road, Chesterfield, MO 63017-3485. Fax No.: (314)206-6850. E-mail: meyejs@stlo.smhs.comAccepted 7 July 1998

Journal of Surgical Oncology 1998;69:212–218

© 1998 Wiley-Liss, Inc.

sentinel nodes. Guiliano et al. [7] found only microme-tastasis in 26 of 68 (38%) of patients with sentinel nodebiopsy plus axillary dissection versus 4 of 39 (10%) ofaxillary dissections without sentinel node biopsy. Inves-tigators listed in Table I reported finding 1.2–2.2 sentinelnodes per patient on the average. As many as eight havebeen found in a given patient [8]. Pathologic examinationhas included keratin staining in some studies (Table I).Precise details of pathologic examination often are lack-ing in published reports, and we have found no analysisof probability of finding metastases of various sizes.

The rationale for sentinel node biopsy includes avoid-ance of axillary dissection when metastases are not pre-sent, cost-saving, and increased sensitivity of detectionof lymph node metastasis. High sensitivity of sentinelnode biopsy for detection of axillary metastasis justifiesavoidance of axillary dissection when the sentinel node isnegative. Alizraki et al. [9] have estimated that sentinelnode examination can save about 80% of lymph nodestaging costs and the morbidity of unnecessary node dis-section. Very thorough examination of the sentinel nodebiopsy is feasible using subserial sectioning because theentire specimen usually can be placed in one or twoparaffin blocks.

METHODS AND RESULTSGeometric Analysis

The simplest model presumes that a lymph node con-tains only one metastasis, and that the metastasis isspherical. Assuming that metastases form in lymph nodesonly if a malignant cell or a group of cells arrive withinthe node, if the metastasis is in the node, its center will bein the node. The center can be anywhere in the node. Thenode may be embedded intact or sliced into two or moresegments. If the slices are 2 mm thick, they will be em-beddeden face.The center of a single metastasis will bein one of the slices. Probability of intercepting the me-tastasis in a microsection can be calculated if the place-ment of the metastasis within the lymph node is randomand the metastasis is spherical. For example, if the thick-ness is broken up into regular intervals of microsections(three totaling 0.015 mm) and unstudied areas (a ribbonof 47 consecutive discarded sections, total4 0.235 mm),eight such intervals will fit into a 2-mm-thick slice (Fig.1). The studied thickness of the slice will be 8 × 0.015mm 4 0.120 mm, and the unstudied thickness will be 4× 0.235 mm4 1.880 mm. The latter will be broken upinto eight segments of 0.235 mm. A 0.1-mm spherical

TABLE I. Sentinel Lymph Node Biopsy for Breast Carcinoma

Lead author(referencecitation) Methoda Nb

Succc

%Posd

%Sense

%Specf

%

Node examination

Sectioningg IHCh

Krag et al. [2] TcS 157 76 34.5 95.1 44 1 or 2 sections NoPijpers et al. [3] TcA 37 92 32 100 36 NDm NoVeronesi et al. [4] TcA 163 98 52 97.5 62 Seven levels Nol

Giuliano et al. [7]i Blue dye 162 100 42 NDm NDm 1 or 2 sections Keratin1 sect.

Turner et al. [8] Blue dye 103 NDm 42 99 NDm Three levels KeratinAlbertini et al. [26] TcS + blue dye 62 92 32 100 (67) 33 NDm NoDale et al. [27] Blue dye 21 66 25 100 40 NDm NoGiuliano et al. [28]i Blue dye 174 65.5 32.4 96 62 NDm False-neg.j

Giuliano et al. [29]i Blue dye 107 93.5 100 33 1 or 2 sections KeratinGuenther et al. [30] Blue dye 145 71 19 97 57 Five levels NoMeijer et al. [31] TcA 28 100 32 100 33 NDk NDk

Nieweg et al. [32] Blue dye 22 86 53 100 40 NDk NDk

Papa et al. [33] Vital dye 48 95 NDk 91 90 NDk NDk

aMethod for localizing sentinel nodes. TcS, technecium 99m sulfur colloid; TcA, Tc albumin colloid; blue dye, isosulfan blue (Nieweg et al. usedpatent blue).bNo. of patients studied.cSuccess rate in localizing sentinel node(s).dPrevalence of axillary metastasis.eSensitivity for detection of axillary metastasis in sentinel node(s).fPrevalence of metastases, one or more, in axillary nodes other than sentinel node(s).gPlan for microsectioning of sentinel node(s).hImmunohistochemical studies.iThese reports are from the same institution: patient populations may overlap.jKeratin stain only for false-negative sentinal nodes.kInformation from abstract; full text paper not available.lLast 107 patients were examined by bisecting sentinel node(s), preparing three frozen sections from each half, and an unstated number ofsections from remnant after fixation and paraffin embedding.mND, not described.

Examination of Sentinel Nodes 213

tumor can be placed with the center anywhere within azone 0.05 mm of either end of the 0.235-mm segment,leaving a span of 0.135 mm in which it can be placedwithout being included in a microsection. Counting thethree 5-mm sections that have been subtracted, the totalpossible span in which the center of the sphere could fallis 0.25 mm. Given random placement of the sphere, theprobability of not encountering it in a section will be0.135/0.2504 0.54, and the probability of finding it willbe 0.46. Equation (1) expresses the probability of findingmetastases of diameters less than the unstudied intervalfor various combinations of sampling intervals and num-bers of sections.

P = 1 − @Tint − $~Tms? Nms! + Dt%#/Tint (1)

where P is the probability of finding the tumor in amicrosection, Tint is thickness of the interval into whichthe slice is divided by the repetitive procedure (0.250 mmin the example above), Tms is thickness of the micro-section (0.005 mm), Nms is the number of microsectionsin each interval, and Dt is the diameter of the sphericaltumor. When the diameter of the tumor exceeds the un-studied interval (0.135 mm in the example above), it willalways be discovered. On the basis of these consider-ations, probabilities of detection of spherical metastasesof different diameters are listed for several illustrativeplans (Tables II, III). When residual block thickness ex-ceeds the sectioning interval, as in plan 2, probability offinding the tumor is the mean of the summed probabili-ties of finding it if in the sectioned part of the block,using equation (1), or unsectioned part of the block,where the probability of finding it is zero.

Biologic Considerations Regarding Micrometastases

Known rates of proliferation and growth of breast car-cinomas provide insight into the duration of metastasesof a given size in lymph nodes. Equation (2) expressesrelationships between volume of the tumor, growth con-stant, and the number of generations

V

V0= ekt (2)

where V is the volume when measured at time t, V0 isvolume at time zero, and e is the base of natural loga-rithms. Time t is measured in units equivalent to cellcycle duration. If t is set to 1 and V/V0 to the incrementin volume during one cell cycle duration, the equationcan be solved for the growth constant k by taking loga-rithms. If the tumor actually doubles during one cellcycle (no cell loss), V/V0 4 2 and k4 0.693, the naturallogarithm of 2. If one-half of the cells generated during acell cycle are lost, V/V0 4 1.5 and k4 0.405. If 90% ofcells are lost, V/V0 4 1.1 and k4 0.095. Under thelatter circumstance, cell proliferation is largely ineffec-tive, and despite rapid cell division growth is slow. Thenumber of generations necessary to reach a given volumewhich contains a given number of cells will depend onrate of cell loss (Table IV). For example, if cell lossfactor is zero, a diameter of 10 mm could be achieved in28 generations. But if cell loss factor is 0.9, 205 genera-tions would be required. The latter circumstance wouldgive the tumor much more opportunity for genetic evo-lution than the former because of the larger number ofcell divisions required to reach detectable size.

Comparison of potential doubling time (mean cellcycle time) [10,11] of breast carcinoma with measuredgrowth rates [12] shows that mean cell loss actually av-erages 90% or more during the clinically detectablephase of disease. Measurements of S-phase fraction of450 infiltrating breast carcinomas by bromodeoxyuridinelabeling of tissue slices showed the mean to be 6.4%,median 3.9% [10]. This is equivalent to a median cellcycle time of 14 days according to Steel’s formula (equa-tion 3) if the median S-phase duration is 18 h [13]. Good-son and associates recorded slightly higher results with invivo bromodeoxyuridine labeling of relatively largebreast carcinomas [14]. The difference is accounted byincrease of mean S-phase fraction with increasing tumorsize [10]. Actual measured growth rates of breast carci-nomas by Spratt and associates indicate a median dou-bling time of 260 days for 448 patients at diagnosis [12].The large disparity between cell cycle time, which isequivalent to potential tumor doubling time, and mea-sured doubling time indicates a high rate of cell loss frombreast carcinoma. Inserting the median values intoSteel’s formula for cell loss [13] would yield a cell lossrate of 95%. High rates of cell loss seem rational whenconsidered in light of rapid proliferative rates of manybreast carcinomas and inherently slow proliferative rateof stromal tissues and their vascular support, which aredependent on cytokines produced by the breast carci-noma cells [15,16]. Evidence for deceleratory growth ofbreast carcinoma has been cited [12], although its actualdemonstration by direct measurement is lacking. Thus,rate of cell loss may be lower when the neoplasm issmall, but long latency of breast carcinoma metastasisbefore clinical presentation argues against more rapid

Fig. 1. Diagram of plans 1, 2, 4, 6, and 7. Tissue blocks are repre-sented on edge at one-fourth height relative to thickness. Scale 13:1;actual block thickness is 2 mm. Dots, metastatic tumors 0.2 mm indiameter. If a tumor does not fall on a vertical line representing amicrosection, it remains undetected.

214 Meyer

growth of micrometastases. Otherwise relapse years andeven decades after initial treatment could not be ex-plained other than by invoking the mysterious concept ofdormancy [17]). For purpose of modeling (vide infra),we chose to include cell loss rates that are actually muchlower than directly measured in breast carcinomas be-cause of uncertainty about these rates in tumors of sub-clinical sizes.

If one accepts the concept that cell loss rates are highfrom the initial stages of breast carcinoma metastaticdevelopment, then the number of generations requiredfor the metastases to reach even small size is impressiveand argues that these metastases have been present formonths, even years. Beginning with the paper by Saphirand Amromin [18], retrospective studies in which lymphnodes have been subserially sectioned have shown a rateof metastasis close to 20% in axillae deemed negativefrom the results of routine histologic study [19]. Earlystudies characterized by small numbers of patients andshort followup commonly failed to demonstrate the prog-nostic effect of micrometastases. Dowlatshahi and asso-ciates [19] reviewed a large number of studies and foundthat significant survival differences commonly were

demonstrated in studies with at least 147 subjects (ninestudies) and follow-up of at least 6 years (one additionalstudy). Within the context of a randomized therapeutictrial to which 921 node-negative patients were acces-sioned, micrometastases were associated with decreasedrelapse-free survival (P 4 0.0008) and decreased sur-vival (P 4 0.0009) at 6 years of observation [20,21].Intuitively, metastases from small carcinomas are likelyto be small in comparison to those from large primarycarcinomas, making detection more difficult. Low prob-ability of detecting small lymph nodal metastasis by tra-ditional examination of one section of each lymph noderaises doubt about reported low rates of positive axillarynodes related to small breast carcinomas [22]. If evensmall tumor deposits have high rates of cell loss, metas-tases could grow slowly enough to remain undetectableby conventional techniques for prolonged periods. TableIV lists the number of cell generation times necessary fortumors to reach a given size when differing rates of cellloss apply. Time elapsed from one mitosis to the next isa generation. It can be approximately calculated fromSteel’s formula if the DNA labeling index (LI) and du-ration of DNA synthesis (Ts) are known. The LI is theproportion of nuclei in S-phase measured by uptake oftritiated thymidine or bromodeoxyuridine:

TG = kTs

LI(3)

where K is a constant accounting for the position of DNAsynthesis in the cell cycle of an exponentially growingpopulation and a reasonable estimate is 0.75 [13]. A rea-sonable approximation of Ts is 18 h from studies on avariety of human cancers [11]. Growth rates have notbeen measured directly for human tumors of subclinicalsize. Measurement of serum prostate-specific antigen(PSA) has shown long doubling times for subclinicalmetastases similar to doubling times measured for clini-cally evident prostate cancer [23], suggesting that, atleast for this type of cancer, small lesions do not possess

TABLE II. Several Plans for Examination of Sentinel Nodes

Plan Description

Cyclelength(mm)

No. ofcycles

No. ofmicrosections Discarded

ribbon length,sections (mm)

Residualtissue (mm)H&Ea Immunob

1 One level 2 1 2 1 0 1.9802 Three levels 0.5 2 4 2 96 0.9853 Two levels 1 2 4 2 196 0.9854 Three levels 1 3 6 3 197 0.0005 Four levels 0.5 4 8 4 97 0.4856 Eight levels 0.25 8 16 8 47 0.2357 Sixteen levels 0.125 16 32 16 22 0.110

aH&E, hematoxylin and eosin.bImmuno, immunostaining.

TABLE III. Probability of Finding Tumor When Node Is Slicedat 2-mm Intervals and Totally Embedded for Microsectioning

Tumordiameter(mm)

Probability of detecting tumor

Plan1

Plan2

Plan3

Plan4

Plan5

Plan6

Plan7

2.0 1 1 1 1 1 1 11.0 0.51 1 1 1 1 1 10.50 0.26 0.76 0.51 0.52 1 1 10.25 0.12 0.40 0.26 0.27 0.53 1 10.20 0.11 0.32 0.23 0.23 0.43 0.86 10.15 0.08 0.25 0.16 0.17 0.33 0.66 10.10 0.06 0.17 0.11 0.12 0.23 0.46 0.920.075 0.05 0.13 0.09 0.10 0.18 0.36 0.720.05 0.03 0.10 0.06 0.07 0.13 0.26 0.520.025 0.02 0.06 0.04 0.05 0.08 0.16 0.320.015 0.02 0.04 0.03 0.04 0.06 0.12 0.240.01 0.01 0.04 0.03 0.03 0.05 0.08 0.20

Examination of Sentinel Nodes 215

more capacity for rapid growth than do large lesions. Itmay then be of some interest to consider how long meta-static tumors of various sizes have been present given anassumption of constant cell generation time and constantrate of cell loss over time (Table V). Given a cell TG of15 days and cell loss factor of 0.9, which are reasonablevalues for breast carcinoma based on available studies, a0.1-mm-diameter metastasis would have been present forapproximately 2.4 years. Even with an extremely lowcell loss factor of zero, it would have been present for 4months and a 0.5-mm metastasis for nearly 8 months.

DISCUSSION

We appear to be at the threshold of an era in whichsentinel lymph node biopsy largely will replace nodeclearance dissection for cancer staging. The studies re-viewed above indicate that this technique is highly sen-sitive for selection of lymph nodes bearing metastases inthe axilla, and similar results have been obtained formalignant melanoma [5,24,25]. By detection and selec-tive biopsy of nodes in the primary lymphatic drainage oftumors, we have opportunity for intensive study of mi-crometastases and their significance in prognosis andtherapy. To grasp this opportunity, we need to examinethe biopsy specimens by systematic techniques that pro-vide known power for detection of metastases of differ-ent sizes. This requires a disciplined approach in thepathology laboratory. Power of detection should be bal-anced against cost, but clearly one or two hematoxylinand eosin (H&E)-stained sections are inadequate for ex-amination of sentinel nodes.

The analysis presented in this paper suggests that pro-cedures can be designed that will find metastases of des-ignated sizes with definable probabilities. Our treatmentassumes random distribution of spherical metastaseswithin lymph nodes. Neither of these assumptions is en-tirely correct. Metastases tend to form in the capsule orsubcapsular regions of lymph nodes. However, using thisknowledge to design histopathological approachespresents spatial difficulties, and we have not attempted to

address this consideration. Shape is dependent on con-straints of surrounding tissue and microenvironmentalconditions influencing growth. In addition, progressionof genetic abnormalities could lead to growth ratechanges focally in a metastasis with resultant distortionof shape. For these reasons, the calculations presented inthis paper are approximations, but they are useful in de-veloping rational designs for histopathologic examina-tion.

The most cost-effective approach will be stepwise.Lymph nodes with grossly visible metastases do not re-quire special treatment. If grossly negative, the first stepwould be one or two microsections from the face of theblock. If negative for metastasis, a systematic serial sec-tioning approach will follow. The procedure to be chosencan be based on the minimum size metastasis one wishesto detect with a high degree of probability. Currently nodata exist to specify what size this should be. Thereforechoice may be driven chiefly by considerations of laborand economics. A 0.25-mm-diameter metastasis can befound with theoretical certainty by plan 6 (Tables II, III),which requires examination of a minimum of 16 micro-sections (or 8 if only one section at each level is studied).This procedure would identify nearly one-half of metas-tases 0.10 mm in diameter and would appear to be prac-tical. The number of microsections prepared by the tissuelaboratory would be comparable to mastectomy and ax-illary node dissection or radical prostatectomy speci-mens, for example, and viewing time by the pathologistwould also be comparable to these specimens. Plan 7,requiring examination of at least 32 (or minimally 16)microsections, would have power to detect virtually allmetastases of 0.15-mm diameter, but would probably betoo tedious for routine application. Either plan could betruncated pending examination of microsections frominitial levels, and deeper levels could be avoided if thetumor is found. However, in order to record more accu-rate dimensions of metastases, the serial sectioning pro-cedure should be carried out to completion. Addition ofimmunohistochemical examination would entail consid-

TABLE IV. Number of Cells and Generations by Different Tumor Sizes

Diameter(mm)

Volume(mm3) No. of cells

No. of generations from one cell with different rates of cell loss

Cell loss4 0 Cell loss4 0.5 Cell loss4 0.9

10 524 296 × 106 28.1 48.1 2055 65.4 37 × 106 25.1 43 1832 4.18 2.4 × 106 21.2 36.2 1551 0.52 294 × 103 18.2 31.1 1360.5 0.065 37 × 103 15.2 25.9 1110.25 0.0082 4,629 12.2 20.8 890.2 0.0042 2,370 11.2 19.2 820.1 0.00052 296 8.2 14 600.05 65 × 10−6 37 5.2 8.9 380.025 8.2 × 10−6 5 2.3 3.9 170.015 1.8 × 10−6 1 0 0 0

216 Meyer

erable increases in expenditure. No consensus on the roleof immunohistochemistry in identifying metastasis hasbeen reached [19]. We believe it should be used routinelywhen the primary lesion is infiltrating lobular carcinomawhich can produce metastases difficult to detect by rou-tine examination. An estimate of histotechnologist’s timefor the sectioning procedure in plan 6 is 15 min perblock. Total costs of thorough systematic study of senti-nel nodes should be seen against a perspective of theimportant light it can shed on breast carcinoma behaviorwhen the metastasizing phenotype is revealed in the mi-crosection. Appropriate reimbursement codes for thistype of sentinel node examination have not yet been de-vised.

Systematic subserial sectioning sentinel node exami-nation can be likened to transaxial computerized imagingof the liver wherein the organ is visualized not by one ortwo capriciously selected cuts, but at regular intervals ofknown width. This model is a necessary one for patholo-gists to follow in order to control power of detection ofmetastases and optimize the histopathologic examina-tion.

REFERENCES

1. Jasinski RW, Smith DC, Chase DR, Field FI: Angiographic pre-operative bowel segment localization using methylene blue, iso-sulfan flue, and fluorescein. Invest Radiol 1987;22:462–466.

2. Krag DN, Ashikaga T, Harlow SP, Weaver D: Development ofsentinel node targeting technique in breast cancer patients. BreastJ 1998;4:67–74.

3. Pijpers R, Meijer S, Hoekstra OS, et al.: Impact of lymphoscin-tigraphy on sentinel node identification with technetium-99m-colloidal albumin in breast cancer. J Nucl Med 1997;38:366–368.

4. Veronesi U, Paganelli G, Galimberti V, et al.: Sentinel-node bi-opsy to avoid axillary dissection in breast cancer with clinicallynegative lymph-nodes. Lancet 1997;349:1864–1867.

5. Wells KE, Rapaport DP, Cruse CW, et al.: Sentinel lymph nodebiopsy in melanoma of the head and neck. Plastic Reconstr Surg1997;100:591–594.

6. Gulec SA, Moffat FL, Carroll RG, Krag DN: Gamma probeguided sentinel node biopsy in breast cancer. Q J Nucl Med 1997;41:251–261.

7. Giuliano AE, Dale PS, Turner RR, et al.: Improved axillary stag-ing of breast cancer with sentinel lymphadenectomy. Ann Surg1995;222:394–399.

8. Turner RR, Ollila DW, Krasne DL, Giuliano AD: Histopathologicvalidation of the sentinel lymph node hypothesis for breast carci-noma. Ann Surg 1997;226:271–276.

9. Alazraki NP, Eshima D, Eshima LA, et al.: Lymphoscintigraphy,the sentinel node concept, and intraoperative gamma probe inmelanoma, breast cancer, and other potential cancers. Semin NuclMed 1997;27:55–67.

10. Meyer JS, Koehm SL, Hughes JM, et al.: Bromodeoxyuridinelabeling for S-phase measurement in breast carcinoma. Cancer1993;71:3531–3540.

11. Wilson GD, McNally NJ, Dische S, et al.: Measurement of cellkinetics in human tumours in vivo using bromodeoxyuridine in-corporation and flow cytometry. Br J Cancer 1988;58:423–431.

12. Spratt JA, von Fournier D, Spratt JS, Weber EE: Deceleratorygrowth and human breast cancer. Cancer 1993;71:2013–2019.

13. Steel GG: ‘‘Growth Kinetics of Tumours.’’ Oxford: ClarendonPress, 1977:66–75.

14. Goodson WH, Ljung BM, Waldman FM, et al.: In vivo measure-ment of breast cancer growth rate. Arch Surg 1991;126:1220–1223.

15. Folkman J: The influence of angiogenesis research on man age-ment of patients with breast cancer. Breast Cancer Res Treat 1995;36:109–118.

16. Guidi AJ, Schnitt SJ, Fischer L et al.: Vascular permeability factor(vascular endothelial growth factor) expression and angiogenesisin patients with ductal carcinoma in situ of the breast. Cancer1997;80:1945–1953.

17. Brinkley D, Haybittle JL: The curability of breast cancer. Lancet1975;2:95–97.

18. Saphir O, Amromin GD: Obscure axillary lymph node metastasesin carcinoma of the breast. Cancer 1948;1:238–241.

19. Dowlatshahi K, Fan M, Snider HC, Habib FA: Lymph node me-tastases from breast carcinoma. Reviewing the dilemma. Cancer1997;80:1188–1197.

20. Bettelheim R, Price KN, Gelber RD, et al., for the InternationalBreast Carcinoma Study Group: Prognostic importance of occultaxillary lymph node micrometastases from breast cancers. Lancet,1990;335:1565–1568.

21. Neville AM, Price K, Gelber RD, Goldhirsch A: Axillary nodemicrometastases and breast cancer. Lancet 1991;337:1110.

22. Silverstein MJ, Gierson ED, Waisman JR, et al.: Axillary lymphnode dissectin for T1a breast carcinoma. Is it indicated? Cancer1994;73:664–777.

23. McLaren DB, McKenzie M, Duncan G, Pickles T: Prostate spe-cific antigen doubling times and clinical behavior in patients withearly untreated prostate carcinoma. Cancer 1998;82:342–348.

24. Glass LF, Messina JL, Cruse W, et al.: The use of intraoperative

TABLE V. Estimated Duration of Metastasis by Size, TDpot, Cell Loss

Diameter(mm)

Duration of metastasis in years dependent on cell loss factor and generation time (TG)

Cell loss4 0 Cell loss4 0.5 Cell loss4 0.9

TG, days TG, days TG, days3 15 55 3 15 55 3 15 55

10 0.23 1.15 4.23 0.40 1.98 7.25 1.69 8.42 30.15 0.21 1.03 3.78 0.35 1.77 6.48 1.50 7.52 27.62 0.17 0.87 3.2 0.30 5.45 5.45 1.27 6.37 23.41 0.15 0.75 2.74 0.26 12.28 4.69 1.12 5.59 20.50.5 0.12 0.62 2.29 0.21 1.06 3.9 0.91 4.56 16.70.25 0.10 0.5 1.83 0.17 0.85 3.13 0.73 3.65 13.40.2 0.09 0.46 1.67 0.16 0.79 2.89 0.67 3.37 12.40.1 0.07 0.34 1.23 0.12 0.58 2.11 0.49 2.47 9.040.05 0.04 0.21 0.78 0.07 0.37 1.34 0.31 1.56 5.730.025 0.02 0.09 0.35 0.03 0.16 0.59 0.14 0.7 2.560.015 0 0 0 0 0 0 0 0 0

Examination of Sentinel Nodes 217

radiolymphoscintigraphy for sentinel node biopsy in patients withmalignant melanoma. Derm Surg 1996;22:715–720.

25. Leong SP, Steinmetz I, Habib FA, et al.: Optimal elective sentinellymph node dissection in primary malignant melanoma. ArchSurg 1997;132:666–672.

26. Albertini JJ, Lyman GH, Cox C, et al.: Lymphatic mapping andsentinel node biopsy in the patient with breast cancer. JAMA1996;276:1818–1822.

27. Dale PS, Williams JT IV: Axillary staging utilizing selective sen-tinel lymphadenectomy for patients with invasive breast carci-noma. Am Surg 1998;64:28–31.

28. Giuliano AE, Kirgan DM, Guenther JM, Morton DL: Lymphaticmapping and sentinel lymphadenectomy for breast cancer. AnnSurg 1994; 220:391–398.

29. Giuliano AE, Jones RC, Brennan M, Statman R: Sentinel lymph-adenectomy in breast cancer. J Clin Oncol 1997;15:2345–2350.

30. Guenther JM, Krishnamoorthy M, Tan LR: Sentinel lymphade-nectomy for breast cancer in a community managed care setting.Cancer J Sci Am 1997;3:336–340.

31. Meijer S, Collet GJ, Pijpers HJ, et al.: Less axillary dissection dueto sentinel node biopsy in patients with breast carcinoma. NederlTijdschr Geneesk 1996;140:2239–2243.

32. Nieweg OE, Kapteijn BA, Peterse JL, et al.: Identification of thesentinel node in patients with breast carcinoma. Nederl TijdschriftGeneesk 1996;140:2235–2239.

33. Papa MZ, Bersuk D, Koler M, et al:. Identification of sentinel andaxillary node involvement in breast cancer. Harefuah 1997;133:428–430.

218 Meyer